Databases Databases are collections of information; our study repeats a theme: Tell the computer the structure, and it can help you! © 2004, Lawrence Snyder

Why Study Databases? Some of us want to compute, but all of us want information … Much of the archived information is in tables Databases enhance applications, e.g. Web Once you know how to create databases, you can use them to personal advantage Databases introduce interesting ideas How much of your information can be in a table?

Stone Age Databases Before relational databases (the kind we study) there were only “flat files” Structural information was difficult to express All processing of information was “special cased” -- custom programs were needed Information repeated; difficult to combine Changes in format of one file means all programs that ever process that file must be changed … adding ZIP codes E.F. Codd of IBM invented relational databases

Relational Databases Information is stored in tables Tables store information about entities -- things or stuff … keep entities of one kind Entities have characteristics called attributes Tables are tuples (rows or records) of attributes (columns or fields ) Every row must be unique, identified by a key Relationships -- associations among the data values are stored Table structure = schema Table contents = instance

A Table Tables have names, attributes, tuples Example: IDnumber unique number(Key) Lasttext person’s last name Firsttext person’s first name Hiredate first day on job Addrtext street address Instance Schema

Redundancy Is Very Bad Not every assembly of tables is a good database -- repeating data is bad Replicated data can differ in its different locations, e.g. multiple addresses can differ Inconsistent data is worse than no data Keep a single copy of any data, and if it is needed in multiple places, associate it with a key, and store key rather than the data

“You can look it up” When looking for information, a single item might be the answer, but a table is more likely “Who is taking CSIS1”? Table of students “Whose mile run time  4:00?” Runner table “Who won 2003 Grammy for ‘Best New Artist?” A table containing only a single row Queries to a DB (set of tables) produces tables

Tables From Tables There are five fundamental operations on tables to create tables: Select -- pick rows from a table Project -- pick columns from a table Union -- combine two tables w/like columns Difference -- remove one table from another Product -- create “all pairs” from two tables Though not primitive “Join” is usually included

Select Operation Select creates a table from the rows of another table meeting a criterion Select_from Example On Hire < 1993

Project Project creates a table from the columns of another table Project Last, First From Example

Union Union (written like addition) combines two tables with same attributes PoliticalUnits = States + Provinces

Difference Difference (written like subtraction) removes 1 table’s rows from another Eastern = States - WestCoast

Product Product (written like multiplication) combines columns and pairs all rows Colors = Blues x Reds Column Rule: If A has x columns, B has y columns, A x B has x+y columns Row Rule: If A has m rows, B has n rows A x B has mn rows There’s divide, too, but forget it

Join Join (written like a bow tie) combines rows (like x ) if common field matches Homes = States Students ><

DB Operations The five DB Operations can create any table from a given set of tables All modern database systems are built on these relational operations Join is not primitive, but can be built from 5 Join, select and project are used most often The operations are not usually used directly, but are used indirectly from other languages SQL, the DB language we learn, is built on basic 5